並列摘要
To improve on outdoor parking lot occupancy detection, we use YOLOv3 to produce a model with excellent accuracy based on the data set of Amato, et al. [1]. Additionally, the cost of equipment and time can be reduced. We analyze various experiment results to determine the effects of environmental and situational factors. We further collect a small localized outdoor parking lot data set in Taiwan, the results of the localized experiments confirm the applicability of transfer learning.